KR102138979B1 - Lane-based Probabilistic Surrounding Vehicle Motion Prediction and its Application for Longitudinal Control - Google Patents

Abstract

A method and apparatus for predicting lane-based probabilistic surrounding vehicle behavior and a longitudinal control method using the same are presented. The lane-based probabilistic neighboring vehicle behavior prediction and longitudinal control method using the method proposed in the present invention include obtaining surrounding vehicle information through a sensor and predicting a target lane of the surrounding vehicle using the acquired surrounding vehicle information. Computing the probability of collision using the target lane and path of the surrounding vehicle considering the step of performing driving route prediction using the acquired surrounding vehicle information for each target lane and the future uncertainty. And proceeding with direction control.

Description

Lane-based Probabilistic Surrounding Vehicle Motion Prediction and its Application for Longitudinal Control

The present invention relates to a lane-based probabilistic neighboring vehicle behavior prediction and a longitudinal control method and apparatus using the same. More specifically, the present invention relates to a stochastic prediction of the behavior of a surrounding vehicle for an autonomous vehicle and a driver assistance system and a longitudinal control method using the same.

As research on intelligent vehicle technologies such as autonomous driving and Driver Assistance Systems (DAS) has been actively conducted, research on predicting behavior of neighboring vehicles is becoming important for securing stability and driving route planning for autonomous vehicles. However, due to the low prediction accuracy of the existing prediction technologies, the safe driving and collision avoidance technologies of autonomous vehicles are also not reliable.

In order to solve this problem, probabilistic prediction in which uncertainty is considered for the behavior of the surrounding vehicle is essential, and an algorithm for probabilistically predicting the target lane and path of the surrounding vehicle using an artificial neural network is required.

In addition, most prediction algorithms fail to properly apply predictions about the behavior of nearby vehicles as stochastic results. Therefore, there is a need for a longitudinal vehicle control algorithm that can calculate a collision avoidance probability by using a probabilistic prediction result value for surrounding vehicles in which uncertainty is considered.

The technical problem to be achieved by the present invention is to provide an algorithm for predicting a target lane and a path of a neighboring vehicle by using a probabilistic behavior prediction, ie, an artificial neural network, in consideration of uncertainty regarding the behavior of the surrounding vehicle. In addition, we propose a longitudinal vehicle control algorithm that can calculate the collision avoidance probability by using the probabilistic predicted results of the surrounding vehicles, where uncertainty is considered.

In one aspect, the lane-based probabilistic neighboring vehicle behavior prediction and longitudinal control method using the method proposed in the present invention include obtaining surrounding vehicle information through a sensor, and using the acquired surrounding vehicle information to obtain a target of the surrounding vehicle Predicting the lane, calculating the driving route prediction using the acquired neighboring vehicle information for each target lane, and calculating the probability of collision using the target lane and route of the surrounding vehicle considering future uncertainty. And performing longitudinal control for collision avoidance.

The step of predicting the target lane of the neighboring vehicle through the artificial neural network structure using the acquired neighboring vehicle information includes an artificial neural network structure having current and past time series lateral information and road information of the surrounding vehicle as input values for a predetermined time. And output the probability for the target lane as an output value.

In addition to the artificial neural network structure, a method of predicting the target lane may use an interacting multiple model or a probability model such as Markov Chain and Gaussian distribution.

The step of performing the driving path prediction and the probability transformation through the artificial neural network structure uses an artificial neural network structure having a longitudinal/lateral position as an input value using a probability for a target lane, and outputs a longitudinal/lateral position. And outputting the output value in the longitudinal/transverse position to probability, finally outputting the longitudinal/transverse path and probability distribution as output values.

As a method of predicting the driving route, an artificial neural network structure or a polynomial fitting may be used. For example, the prediction of a driving route using a polynomial fitting may generate a route in a polynomial form with the current position of the lane and the center line of the target lane when the target lane is determined.

The step of performing the longitudinal control for collision avoidance using the probabilistic result of the collision probability uses the cost function that minimizes the difference between the longitudinal target velocity and the current velocity and the target acceleration to determine the collision probability value. Optimal longitudinal control is performed so that the cost function is minimized without exceeding.

By using the probabilistic behavior prediction of the behavior of surrounding vehicles, it is possible to control the longitudinal safety distance between the host vehicle and the surrounding vehicle by using the opportunity constraint type, and it is possible to control the longitudinal direction considering the uncertainty of the collision. It is possible to reflect the driving propensity of the own vehicle driver.

In another aspect, the lane-based probabilistic neighboring vehicle behavior prediction proposed by the present invention and the longitudinal control device using the same are a sensor for acquiring surrounding vehicle information and a target lane of the surrounding vehicle using the acquired surrounding vehicle information Prediction unit for predicting, Probability calculation unit for predicting the driving route using the information about the surrounding vehicle acquired for each target lane, and the probability for the possibility of collision using the target lane and path of the surrounding vehicle considering the future uncertainty And a longitudinal control unit for calculating and performing longitudinal control for collision avoidance.

In another aspect, the method for predicting lane-based probabilistic surrounding vehicle behavior and a longitudinal control method using the method proposed by the present invention include obtaining surrounding vehicle information through a sensor, and using the acquired surrounding vehicle information, an artificial neural network Predicting target lanes of the surrounding vehicles through the structure, performing driving route prediction using the acquired surrounding vehicle information for each target lane, and longitudinal targets using the predicted target vehicle lanes and driving routes And calculating a cost function to minimize the difference between the speed and the current speed and the target acceleration, and performing optimal longitudinal control so that the cost function is minimized without a collision probability exceeding a predetermined value.

In another aspect, the lane-based probabilistic neighboring vehicle behavior prediction proposed by the present invention and the longitudinal control device using the predicted vehicle through the artificial neural network structure using a sensor that acquires the surrounding vehicle information and the acquired surrounding vehicle information A prediction unit for predicting a target lane of a nearby vehicle, a probability calculation unit for performing a driving route prediction using information of surrounding vehicles obtained for each target lane, and a longitudinal target using a predicted target vehicle's target lane and a driving route It includes a longitudinal control unit that calculates a cost function that minimizes the difference between the speed and the current speed and the target acceleration, and performs optimal longitudinal control so that the cost function is minimized without a collision probability exceeding a predetermined value.

According to embodiments of the present invention, a predictive behavior prediction in which uncertainty is considered for a behavior of a surrounding vehicle, that is, a target lane and a path of a surrounding vehicle can be predicted using an artificial neural network, and a surrounding vehicle in which uncertainty is considered It is possible to calculate the probability of a collision possibility using the target lane and path of and to control the longitudinal direction for collision avoidance.

1 is a flowchart illustrating a lane-based probabilistic neighboring vehicle behavior prediction according to an embodiment of the present invention and a longitudinal control method using the same.
2 is a view for explaining a probability behavior prediction for a lane-based neighboring vehicle using an artificial neural network according to an embodiment of the present invention.
3 is a diagram for explaining the structure of an artificial neural network for target lane prediction and the target lane and probability of a surrounding vehicle according to an embodiment of the present invention.
4 is a diagram illustrating an artificial neural network structure for predicting a longitudinal and transverse driving path according to an embodiment of the present invention.
5 is a diagram illustrating a process of predicting a driving route in consideration of uncertainty according to an embodiment of the present invention.
6 is a view for explaining longitudinal control using model prediction control according to an embodiment of the present invention.
7 is a formula showing definition and constraints on a cost function of model prediction control according to an embodiment of the present invention.
8 is a view showing a safe situation and a dispatch situation according to an embodiment of the present invention.
9 is a diagram illustrating a structure of a lane-based probabilistic peripheral vehicle behavior prediction and a longitudinal control device using the same according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart illustrating a lane-based probabilistic neighboring vehicle behavior prediction according to an embodiment of the present invention and a longitudinal control method using the same.

The proposed lane-based probabilistic neighboring vehicle behavior prediction and longitudinal control method using the method include obtaining nearby vehicle information through a sensor (110), and predicting a target lane of the surrounding vehicle using the acquired nearby vehicle information (120), a step of performing a driving route prediction using the acquired surrounding vehicle information for each target lane (130), calculating the probability of a collision possibility using the target lane and path of the surrounding vehicle in consideration of future uncertainty And proceeding to the longitudinal control for collision avoidance (140).

In step 110, surrounding vehicle information is acquired through a sensor.

In step 120, the target lane of the neighboring vehicle is predicted using the acquired surrounding vehicle information. An artificial neural network structure using current and past time series lateral information and road information of surrounding vehicles for a predetermined time as input values may be used. Then, the probability for the target lane is output as an output value. In addition to the artificial neural network structure, a method of predicting the target lane may use an interacting multiple model or a probability model such as Markov Chain and Gaussian distribution.

In step 130, the driving route prediction is performed using the acquired surrounding vehicle information for each target lane. An artificial neural network structure using current and past time series longitudinal/transverse positions or speeds of surrounding vehicles as input values for a predetermined time is used, and longitudinal/transverse positions are output as output values. In this process, different artificial neural networks are used for the longitudinal and transverse directions. In addition, through the process of converting the output value to probability, finally, the probability for the longitudinal/transverse position is output as the output value. As a method of predicting the driving route, a driving route prediction using an artificial neural network structure or a polynomial equation may be used. For example, the prediction of a driving route using a polynomial fitting may generate a route in a polynomial form with the current position of the lane and the center line of the target lane when the target lane is determined.

In step 140, a probability of a collision possibility is calculated using a target lane and a path of a surrounding vehicle in which future uncertainty is considered, and longitudinal control for collision avoidance is performed. By using a cost function that minimizes the difference between the longitudinal target speed and the current speed and the target acceleration, optimal longitudinal control is performed so that the cost function is minimized without exceeding a predetermined value.

By using the probabilistic behavior prediction for the behavior of the surrounding vehicle, it is possible to control the longitudinal safety distance between the host vehicle and the surrounding vehicle using the opportunity constraint type, and thus, the longitudinal control considering the uncertainty of collision is possible. In addition, it is possible to reflect the propensity of driving of the own vehicle driver through parameter adjustment.

2 is a view for explaining a probability behavior prediction for a lane-based neighboring vehicle using an artificial neural network according to an embodiment of the present invention.

As shown in FIG. 2, an algorithm is proposed that predicts a target lane and a driving path of a neighboring vehicle through an artificial neural network and changes the result to a probabilistic result value by using information about surrounding vehicles acquired through a sensor in order.

Acquire the surrounding vehicle information 210 through the sensor, and predict the target lane of the surrounding vehicle 220 using the acquired surrounding vehicle information. The driving route prediction 230 is performed using the acquired surrounding vehicle information for each target lane. At this time, an artificial neural network structure using current and past time series positions and road information of nearby vehicles for a predetermined time may be used. In addition, it is possible to use an artificial neural network structure that uses current and past longitudinal positions of surrounding vehicles as time series input values for a predetermined time and future longitudinal positions as output values for a predetermined time. Subsequently, the probability of a collision possibility may be calculated using the target lane and the path of the surrounding vehicle in consideration of future uncertainty, and longitudinal control for collision avoidance may be performed 240.

3 is a diagram for explaining the structure of an artificial neural network for target lane prediction and the target lane and probability of a surrounding vehicle according to an embodiment of the present invention.

As shown in FIG. 3, an artificial neural network structure using current and past time series positions and load information of surrounding vehicles for a predetermined time may be used. According to an embodiment of the present invention, an artificial neural network structure using current and past time series position and load information for 3 seconds of a surrounding vehicle as input values and a probability for a target lane of a surrounding vehicle as a result value is used.

4 is a diagram illustrating an artificial neural network structure for predicting a longitudinal and transverse driving path according to an embodiment of the present invention.

As shown in FIG. 4, an artificial neural network structure may be used in which a current and past longitudinal position of a surrounding vehicle is a time series input value for a predetermined time and a future longitudinal position is an output value for a predetermined time. According to an embodiment of the present invention, a lane change and lane maintenance of a neighboring vehicle are predicted through a probability for a target lane, and current and past lateral positions for 3 seconds for each case of lane change and maintenance as shown in FIG. 4 are shown. It uses an artificial neural network structure that takes the time series input value and outputs the lateral position of the future for 5 seconds.

5 is a diagram illustrating a process of predicting a driving route in consideration of uncertainty according to an embodiment of the present invention.

Through the process of predicting the path of the surrounding vehicle through the predicted position result value, as shown in FIG. 5, a probabilistic path prediction result having a certain reliability may be derived through a process of probabilistically converting the result value for the driving path.

Predict the lane change and maintenance of nearby vehicles through stochastic results for the target lane. In addition, for each case of changing and maintaining lanes of surrounding vehicles, an artificial neural network structure using current and past lateral positions as time series input values for a predetermined time and future lateral positions as output values for a predetermined time may be used. Can. In addition, it is possible to predict the path of the surrounding vehicle through the predicted longitudinal position result value, and convert the result value for the driving path into a stochastic result value.

6 is a view for explaining longitudinal control using model prediction control according to an embodiment of the present invention.

After predicting the probabilistic behavior of the neighboring vehicle using the artificial neural network technique, collision avoidance longitudinal control may be performed through model predictive control using a probability for a collision possibility based on this, as shown in FIG. 6.

After predicting the probabilistic behavior of the surrounding vehicle through the artificial neural network structure, longitudinal control for collision avoidance is performed using the probability of collision probability. By using a cost function that minimizes the difference between the longitudinal target speed and the current speed and the target acceleration, optimal longitudinal control is performed so that the cost function is minimized without exceeding a predetermined value.

By using the probabilistic behavior prediction for the behavior of the surrounding vehicle, it is possible to control the longitudinal safety distance between the host vehicle and the surrounding vehicle using the opportunity constraint type, and thus, the longitudinal control considering the uncertainty of collision is possible. In addition, it is possible to reflect the propensity of driving of the own vehicle driver through parameter adjustment.

7 is a formula showing definitions and constraints on a cost function of model prediction control according to an embodiment of the present invention.

Equation (1) of FIG. 7 is a cost function that minimizes the difference between the longitudinal target speed and the current speed and the target acceleration, and performs optimal control in the direction in which the cost function is minimized.

Equations (2), (3), (4), and (5) of FIG. 7 may use acceleration range, acceleration change amount, and probability of collision in consideration of physical limitations of the vehicle as constraints in minimizing the cost function.

8 is a view showing a safe situation and a dispatch situation according to an embodiment of the present invention.

In the case of Equation (5) of FIG. 7, since the result of the prediction of the behavior of the surrounding vehicle is probabilistic as shown in FIG. 8, the longitudinal safety distance limitation between the host vehicle and the surrounding vehicle is used for collision using the opportunity constraint form. Longitudinal control considering uncertainty is possible, and the driver's propensity to drive can be reflected through parameter adjustment.

9 is a diagram illustrating a structure of a lane-based probabilistic peripheral vehicle behavior prediction and a longitudinal control device using the same according to an embodiment of the present invention.

The proposed lane-based probabilistic neighboring vehicle behavior prediction and the longitudinal control device 900 using the same include a sensor 910, a prediction unit 920, a probability calculation unit 930, and a longitudinal control unit 940.

The sensor 910 acquires surrounding vehicle information.

The prediction unit 920 predicts a target lane of the surrounding vehicle using the acquired surrounding vehicle information. An artificial neural network structure using current and past time series lateral information and road information of surrounding vehicles for a predetermined time as input values may be used. Then, the probability for the target lane is output as an output value.

The probability calculation unit 930 performs driving route prediction by using the surrounding vehicle information obtained for each target lane. Using an artificial neural network structure that uses the current and past time series longitudinal/transverse position or speed of the surrounding vehicle as an input value for a predetermined time using a probability for the target lane, and outputs the longitudinal/transverse position as an output value do. In this process, different artificial neural networks are used for the longitudinal and transverse directions. In addition, through the process of converting the output value to probability, finally, the probability for the longitudinal/transverse position is output as the output value.

The longitudinal control unit 940 calculates a probability of a collision possibility using a target lane and a path of a surrounding vehicle in which future uncertainty is considered, and performs longitudinal control for collision avoidance. By using a cost function that minimizes the difference between the longitudinal target speed and the current speed and the target acceleration, optimal longitudinal control is performed so that the cost function is minimized without exceeding a predetermined value.

By using the probabilistic behavior prediction for the behavior of the surrounding vehicle, it is possible to control the longitudinal safety distance between the host vehicle and the surrounding vehicle using the opportunity constraint type, and thus, the longitudinal control considering the uncertainty of collision is possible. In addition, it is possible to reflect the propensity of driving of the own vehicle driver through parameter adjustment.

The proposed lane-based probabilistic neighboring vehicle motion prediction and longitudinal control method and apparatus using the same, the vehicle on the road runs with the regularity of driving along the lane in both highway driving and downtown driving, but existing technologies Since the lane-based model was not used, consideration was not given to this. However, in the present invention, it is possible to predict the behavior of a lane-based neighboring vehicle using the location and speed of the surrounding vehicle or time and location information of the past and present as input values.

Prediction of the behavior of surrounding vehicles through deterministic methods can never have 100% accuracy. Therefore, a probabilistic prediction considering uncertainty is needed.

In predicting the stochastic behavior of an adjacent vehicle using an artificial neural network, the existing NN technique only deterministically estimates only the future location, but the present invention considers uncertainty through stochastic behavior prediction. In addition, accuracy can be further improved by using not only current data but also past and present data as input values.

In most controllers, constraints cannot consider the uncertainty of the prediction result, but in the case of model prediction control using the opportunity constraint, the opportunity constraint can express the possibility of collision using the uncertainty included in the prediction result.

According to embodiments of the present invention, it can be applied to a vehicle autonomous driving technology and a driver assistance system. Due to the low accuracy of prediction of the behavior of surrounding vehicles on the road, artificial neural networks are based on lane-based behavior based on the lane based on the fact that the vehicle must move according to the rules of lane at this point in time when advanced driving systems do not gain trust. By using probabilistic prediction by using the predicted result that includes uncertainty in the future, it can be applied to driver support systems such as SCC, AEB, as well as technology for driving route planning and collision avoidance operation of autonomous vehicles.

In the global auto industry, the development of autonomous driving technology has been actively conducted according to the commercialization plan of autonomous driving. The auto industry sees the commercialization time of autonomous vehicles as 2020, and it can be said that the surrounding vehicle prediction technology corresponding to the core technology of autonomous vehicles will also increase in importance.

Peripheral vehicle prediction is an essential part of the development of autonomous driving technology, and considering the current time when 100% reliability is not secured, probabilistic prediction of the behavior of the surrounding vehicle is expected to be used for all autonomous vehicle development.

As the market for autonomous driving technology is expected to be applied to all autonomous driving technologies by preoccupying core technologies at the time of the growing market, lane-based probabilistic neighboring vehicle behavior prediction proposed by the present invention and longitudinal control method using the same It is expected to play an important role.

In addition, it is expected to affect not only the autonomous driving technology but also the driver assistance system currently in use, which will affect the overall automotive industry by developing the technology.

It is possible to lead the autonomous driving market through the development of core technologies for fully autonomous driving, and it is expected that the position of domestic technology in the economic and industrial aspects will rise dramatically through technology export.

The device described above may be implemented with hardware components, software components, and/or combinations of hardware components and software components. For example, the devices and components described in the embodiments include, for example, processors, controllers, arithmetic logic units (ALUs), digital signal processors (micro signal processors), microcomputers, field programmable arrays (FPAs), It may be implemented using one or more general purpose computers or special purpose computers, such as a programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. The processing device may run an operating system (OS) and one or more software applications running on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of the software. For convenience of understanding, a processing device may be described as one being used, but a person having ordinary skill in the art, the processing device may include a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that may include. For example, the processing device may include a plurality of processors or a processor and a controller. In addition, other processing configurations, such as parallel processors, are possible.

The software may include a computer program, code, instruction, or a combination of one or more of these, and configure the processing device to operate as desired, or process independently or collectively You can command the device. Software and/or data may be interpreted by a processing device or to provide instructions or data to a processing device, of any type of machine, component, physical device, virtual equipment, computer storage medium or device. Can be embodied in The software may be distributed on networked computer systems, and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, or the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and usable by those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, and magnetic media such as floptical disks. -Hardware devices specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language code that can be executed by a computer using an interpreter, etc., as well as machine language codes produced by a compiler.

As described above, although the embodiments have been described by a limited embodiment and drawings, those skilled in the art can make various modifications and variations from the above description. For example, the described techniques are performed in a different order than the described method, and/or the components of the described system, structure, device, circuit, etc. are combined or combined in a different form from the described method, or other components Alternatively, even if replaced or substituted by equivalents, appropriate results can be achieved.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims (16)

Obtaining surrounding vehicle information through a sensor;
Predicting a target lane of the surrounding vehicle using the acquired surrounding vehicle information;
Performing a driving route prediction using the acquired surrounding vehicle information for each target lane; And
Calculating the probability of the possibility of collision using the target lane and the path of the surrounding vehicle considering the future uncertainty and proceeding to the longitudinal control for collision avoidance
Including,
Predicting a target lane of the surrounding vehicle using the acquired surrounding vehicle information,
The artificial neural network structure using current and past time series lateral information and road information of surrounding vehicles as input values for a predetermined time, different artificial neural networks are used for the longitudinal and lateral directions, and the probability for the target lane Output as the output value,
The longitudinal safety distance constraint between the host vehicle and the surrounding vehicle is performed using the opportunistic constraint form to perform the longitudinal control considering the uncertainty of the collision, and reflect the propensity of driving of the own vehicle driver through parameter adjustment.
Prediction of surrounding vehicle behavior and control of own vehicle. delete According to claim 1,
The method for predicting the target lane uses an Interacting Multiple Model or Markov Chain in addition to the artificial neural network structure.
A method for predicting surrounding vehicle behavior and controlling own vehicle. According to claim 1,
The step of performing driving route prediction using the acquired surrounding vehicle information for each target lane may include:
Using an artificial neural network structure that takes the current and past time series longitudinal/transverse position or speed of the surrounding vehicle as an input value for a predetermined time, and outputs the longitudinal/transverse position as an output value
A method for predicting surrounding vehicle behavior and controlling own vehicle. According to claim 4,
The method for predicting the driving route uses a polynomial fitting in addition to the artificial neural network structure.
A method for predicting surrounding vehicle behavior and controlling own vehicle. According to claim 4,
After the process of converting the output value output in the longitudinal/transverse position to probability, finally outputting the probability for the longitudinal/transverse position as the output value
A method for predicting surrounding vehicle behavior and controlling own vehicle. According to claim 1,
The step of calculating the probability of the possibility of collision using the target lane and the path of the surrounding vehicle considering the future uncertainty and proceeding to the longitudinal control for collision avoidance,
Using the cost function that minimizes the difference between the longitudinal target speed and the current speed and the target acceleration, optimal longitudinal control is performed so that the cost function is minimized without exceeding a predetermined value.
A method for predicting surrounding vehicle behavior and controlling own vehicle. The method of claim 7,
By using the probabilistic behavior prediction of the behavior of surrounding vehicles, it is possible to control the longitudinal safety distance between the host vehicle and the surrounding vehicle by using the opportunity constraint type, and it is possible to control the longitudinal direction considering the uncertainty of the collision. It is possible to reflect the propensity of driving of the own vehicle
Prediction of surrounding vehicle behavior and control of own vehicle. A sensor for acquiring surrounding vehicle information;
A prediction unit for predicting a target lane of the surrounding vehicle using the acquired surrounding vehicle information;
A probability calculation unit for performing a driving route prediction using the acquired surrounding vehicle information for each target lane; And
Longitudinal control unit that calculates the probability of a collision possibility using the target lane and path of the surrounding vehicle considering the future uncertainty and performs longitudinal control for collision avoidance
Including,
The Prediction Department,
The artificial neural network structure using current and past time series lateral information and road information of surrounding vehicles as input values for a predetermined time, different artificial neural networks are used for the longitudinal and lateral directions, and the probability for the target lane Output as the output value,
The longitudinal safety distance constraint between the host vehicle and the surrounding vehicles can be controlled in the longitudinal direction considering the uncertainty of collision by using the opportunity constraint type, and the driver's propensity to drive is reflected through parameter adjustment.
A device for predicting surrounding vehicle behavior and controlling own vehicle. delete The method of claim 9,
The probability calculation unit,
Using an artificial neural network structure that takes the current and past time series longitudinal/transverse position or speed of the surrounding vehicle as an input value for a predetermined time, and outputs the longitudinal/transverse position as an output value
A device for predicting surrounding vehicle behavior and controlling own vehicle. The method of claim 9,
The longitudinal control unit,
Using the cost function that minimizes the difference between the longitudinal target speed and the current speed and the target acceleration, optimal longitudinal control is performed so that the cost function is minimized without exceeding a predetermined value.
A device for predicting surrounding vehicle behavior and controlling own vehicle. delete delete delete delete

Images